In getting my notes together for a root install that I want to try, using the OpenZFS manual setup, I have 3 questions to clarify some things.
https://openzfs.github.io/openzfs-do...0on%20ZFS.html
There will be 2 NVMe's - each 1 TB. the 2nd will be a Mirror of the 1st.
On the Primary NVMe will be a root install, per the openzfs setup, plus an additional partition & pool for data.
#1 - I read in various places that you "should" give ZFS the full disk, and other resources that say do not. My inclination is before starting anything, to set up a GPT partition table with a single large partition spanning say 97% of the disk, as an example, and then starting the installations on this "part1". Is this necessary? I also have read that when you do give ZFS the full disk, that zfs will take care creating a starting and ending point on the disk. Not sure which way to go here.
#2 - The setup talks about creating the same set of partitions (bpool, rpool, swap, etc) on $DISK2, as well as the primary disk. I thought creating the Mirror relationship between the 2 disks would handle everything on DISK2. If when a disk fails, you replace it, doesn't the resilvering process recreate everything on the new disk? If so, why would the installation require manual partitioning on DISK2 (the mirror)?
#3 - I want to add a data partition & pool on the same NVMe. Is it best to complete the root setup first, then create the data partition & pool?